Template for Comments and Response by the Editor on the Comments Submitted Date: 2005/01/27 Document: ISO/IEC DIS 22537

Template for comments and response by the Editor on the comments submitted Date: 2005/01/27 Document: ISO/IEC DIS 22537

1 2 (3) 4 5 (6) (7)

MB1 Clause No./ Paragraph/ Type Comment (justification for change) by the MB Proposed change by the MB Response Subclause Figure/Table/ of on each comment submitted No./ Note com- Annex (e.g. Table 1) ment2 (e.g. 3.1)

Ecma 9.1.1.1 te It is possible to get XML values whose set of There are four changes necessary to address this Accept. fix mapinfoitem and [[InScopeNamespaces]] is inconsistent with the set of issue: remove notes and unwinding namespaces used by that value. For example, this might the parent chain to find occur when an element is extracted from its parent. (1) In MapInfoItemToXML: inscope namespaces.

While the specification contains many notes of the form Step 6: remove step g and modify the new step g Keep 1, 3 and 4. Add 5 to (was step h) to look like: remove notes and while NOTE: The E4X data model does not enforce the loops that enforce the constraint: g. For each attribute information item a in the [in- constraint in 2. for all x belonging to XML: scope namespaces] property of i, except for the In 9.1.12 specify that the x.[[InScopeNamespaces]] is an improper attribute information item whose [prefix] property is replaced child inherits the superset of equal to “xml” children. x.[[Parent]].[[InScopeNamespaces]]. i. Map a member ns of x.[[InScopeNamespaces]] to a as follows: to allow implementations freedom of representation, it 1. Map ns.prefix to the [prefix] property must also be true that the namespaces used by an XML of a value must always be in the [[InScopeNamespaces]] of 2. Map ns.uri to the [namespace name] that value. property of a

(2) Constrain [[InScopeNamespaes]] of an XML value to include every namespace used by that XML value

(3) Change the algorithm of ToXMLString() to emit a xmlns declaration for the default xml namespace rather than generating an prefix for it

(4) Change the algorithm in [[AddInScopeNamespaces]] to replace the default namespace. Ecma 9.1.1.1, ed There is redundant code that should be shared between Define an abstract procedure to describe the This comment is technical. 9.1.1.3 these two sections. common logic shared between these two clauses. Defer to edition 4 Ecma 9.1.1.2 te ToString is called unnecessarily, which is inefficient and Move steps 3 and 4 to before 1 and 2 Accept can have side effects.

1 MB = Member body (enter the ISO 3166 two-letter country code, e.g. CN for China; comments from the ISO/CS editing unit are identified by **) 2 Type of comment: ge = general te = technical ed = editorial NOTE Columns 1, 2, 4, 5 are compulsory. page 1 of 13 ISO electronic balloting commenting template/version 2001-10 Template for comments and response by the Editor on the comments submitted Date: 2005/01/27 Document: ISO/IEC DIS 22537

1 2 (3) 4 5 (6) (7)

Ecma 9.1.1.2 te Step 6. GetDefaultNamespace is called much earlier than Inline or move down Accept its one use. Ecma 9.1.1.2 ed Step 9. isValueName misspelling of isValidName Replace isValidName with isValueName Accept Ecma 9.1.1.3, ge Throughout the document, the handling of conditional (see comment) Accept 9.2.1.3, and statements is inconsistent. I.e. elsewhere The following pattern where there are extra steps with else-after-return and extra returns: 2 If foo 2(a) If bar, return true 2(b) Else 2(b)(i) blah 2(b)(ii) blah etc. 2(c) Return true

Should be rewritten as

2 If foo 2(a) If !bar 2(a)(i) blah 2(a)(ii) blah etc. 2(b) Return true

Ecma 9.1.1.4 ed The terminology “shift up" vs. “shift down" is unclear. Use the terminology: “shift higher” and “shift lower” Accept instead of “shift up” and “shift down”, respectively. Ecma 9.1.1.13 ed [[AddInScopeNamespaces]] wrongly plural in 3(d)(iii) Replace [[AddInScopeNamespaces]] with Accept [[AddInScopeNamespace]] Ecma 9.2.1.2 te Step 2(c)(ii) sets y.[[Parent]] = r where r is the result of To match insertChildAfter, insertChildBefore, Accept [[ResolveValue]] called on x.[[TargetObject]] in 2(a)(i). prependChild, and setChildren, we should silently

1 MB = Member body (enter the ISO 3166 two-letter country code, e.g. CN for China; comments from the ISO/CS editing unit are identified by **) 2 Type of comment: ge = general te = technical ed = editorial NOTE Columns 1, 2, 4, 5 are compulsory. page 2 of 13 ISO electronic balloting commenting template/version 2001-10 Template for comments and response by the Editor on the comments submitted Date: 2005/01/27 Document: ISO/IEC DIS 22537

1 2 (3) 4 5 (6) (7)

This can result in text parenting text. E.g. do nothing in this case.

var MYXML = new XML(); MYXML.appendChild( new XML("Giants"));

Ecma 9.2.1.2 ed Step 2(c)(vii)(1) could test r instead of y.[[Parent]], since (see comment) Accept we know from 2(c)(ii) that they're identical.

Ecma 9.2.1.2 te Step 2(c)(vii)(3) what is V.[[PropertyName]]? Should be (see comment) Accept [[TargetProperty]] Ecma 9.2.1.2 te Step 2(f)(iv, vi). Replace (iv) and (vi) with these steps: Accept Off-by-one error iv. For j = x.[[Length]]-1 downto i + 1, rename property j of x to ToString(j + c.[[Length]] - 1)

vi. Let x.[[Length]] = x.[[Length]] + c.[[Length]] - 1 Ecma 9.2.1.2 te Step 2(g)(iii)V may not be of type XML, but all index- Thanks to 2(d), we know V is either a string or an Accept named properties x[i] in an XMLList x must be of type XML/XMLList object. If V is a string, call ToXML XML, according to 9.2.1.1 Overview and other places in on it to satisfy the constraint before setting x[i] = V. the spec. Ecma 9.2.1.2 Step te All uses of a.[[Name]] for an attribute a in these sections Need to convert QName into an attribute name Accept 2(e)(i, ii), that pass that QName object to [[Delete]] must pass an before calling [[Delete]] 9.2.1.2 Step AttributeName cloned from a.[[Name]]. The [[Name]] 7(e)(i), internal property is always a QName instance and never 9.2.1.3 Step an AttributeName or AnyName instance. But [[Delete]] 2(b)(ii)(1)(a) will not operate on x.[[Attributes]] when given a QName by these sections, so a child could be wrongly deleted instead of the attribute of the same name. 1 MB = Member body (enter the ISO 3166 two-letter country code, e.g. CN for China; comments from the ISO/CS editing unit are identified by **) 2 Type of comment: ge = general te = technical ed = editorial NOTE Columns 1, 2, 4, 5 are compulsory. page 3 of 13 ISO electronic balloting commenting template/version 2001-10 Template for comments and response by the Editor on the comments submitted Date: 2005/01/27 Document: ISO/IEC DIS 22537

1 2 (3) 4 5 (6) (7)

Ecma 9.2.1.10 ed 2(a) tests for null after testing for a non-null value 2(a) transpose 2nd and 3rd terms in if condition's Accept disjunction Ecma 10.2.1 te Step 3 is premature, given the early returns in steps 4-7. Add s+return_value where returns do not already Accept Unless steps 4-7 should prepend s to their specified do that. return values, which upon testing seems like the right thing! So the errata are that (4-7) do not prepend s to their "Return" results.

Ecma 10.2.1 te Step 11 seems to make a copy of the in-scope Move copying closer to use Accept namespace prematurely. Only if Step 12's "If (namespace.prefix == undefined)" test is true does it need the copy, in order to set namespace.prefix.

Ecma 10.2.1 te Step 12 seems confused: if namespace.prefix is set to an Remove test make 12.b.i 12.b Accept arbitrary prefix not used by any namespace in the union set, then 12(b)'s "If" condition is always true, and we'll always add the newly-prefixed copy of the namespace found in the in-scope namespaces to namespaceDeclarations.

Ecma 10.2.1 ed Step 17(b)(i) typo: [[GetNamspace]] Replace with [[GetNamespace]] Accept

Ecma 10.2.1 te Step 17(e-f): The XML spec say these need Add a step to call EscapeAttributeValue with the Accept. Replace “fo” with “of” EscapeAttributeValue attribute value as the argument before concatenating the result Ecma 10.2.1 ed Step 21(a) Typo: indentLevel is wrongly capitalized. Fix capitalization Accept Ecma 10.2.1 te Step 24(a) It seems a new line character should be Add a step before Step 24(a), which adds line Accept concatenated to s before the indentLevel spaces, to put terminator character to s. the end-tag on its own line.

1 MB = Member body (enter the ISO 3166 two-letter country code, e.g. CN for China; comments from the ISO/CS editing unit are identified by **) 2 Type of comment: ge = general te = technical ed = editorial NOTE Columns 1, 2, 4, 5 are compulsory. page 4 of 13 ISO electronic balloting commenting template/version 2001-10 Template for comments and response by the Editor on the comments submitted Date: 2005/01/27 Document: ISO/IEC DIS 22537

1 2 (3) 4 5 (6) (7)

Ecma 10.2.1.1 te Step 2(b) is unnecessary and contrary to user Remove step 2(b). Reject. Continue to escape expectations because ‘>’ characters do not need to be for parallelism with ‘<’. escaped in element content. Ecma 10.6.1 te Step 1 uses P, should be s. Step 1 excludes any number Change S to p. Change conversion to ToUInt32() Accept values property when it should only exclude unsigned integer valued properties. Ecma 11.1 te Step 3(a)(i) Throw ReferenceError if not found Accept n::x given valid Namespace reference n, @n::b, *, etc. => undefined if not found in scope chain. This goes against Editions 1-3 and the implementations that led to the EcmaScript standard, and it's not good human engineering. Ecma 11.3.2 ed The specification seems to be ambigious as to what Return the string “xml”, instead of “xmllist” Accept should be returned as section 11.3.2 The typeof Operator says in its text When UnaryExpression evaluates to a value of type XMLList, the typeof operator returns the string "xmllist", while the table defining the results says: Type Result XML "xml" XMLList "xml"

Ecma 11.1.4 te The grammar for XML initialisers includes markup that is Replace the grammar for XML initialisers with the Accept. not well-formed XML. A tighter grammar will allow syntax following: errors to be caught while parsing the program, rather than at runtime. XMLElement < XMLName XMLAttributes XMLWhitespaceOpt/> < XMLName XMLAttributes > XMLElementContent

XMLName { Expression } XMLName

XMLAttributes 1 MB = Member body (enter the ISO 3166 two-letter country code, e.g. CN for China; comments from the ISO/CS editing unit are identified by **) 2 Type of comment: ge = general te = technical ed = editorial NOTE Columns 1, 2, 4, 5 are compulsory. page 5 of 13 ISO electronic balloting commenting template/version 2001-10 Template for comments and response by the Editor on the comments submitted Date: 2005/01/27 Document: ISO/IEC DIS 22537

1 2 (3) 4 5 (6) (7)

XMLWhitespace { Expression } XMLAttribute XMLAttributes empty

XMLAttribute XMLWhitespace XMLName XMLWhitespaceOpt = XMLWhitespaceOpt XMLAttributeValue

XMLAttributeValue { Expression } XMLAttributeValue

XMLElementContent { Expression } XMLElementContent XMLMarkup XMLElementContent XMLText XMLElementContent XMLElement XMLElementContent empty

Ecma 11.2.4 ed 6(a, d, e) should use l[i], not x[i]. Replace references to x[i] with l[i] Accept

Ecma 11.5.1 te E4X specs the section 11.5.1 The Abstract Equality That should be changed to Accept Comparison Algorithm c. If Type(x) is Object and states: x.[[Class]] == "Namespace" and y.[[Class]] == "Namespace", ... return the results of the comparison 3. If Type(x) is the same as Type(y) x.uri == y.uri ... c. If Type(x) is Object and A similar treatment should be applied to QName x.[[Class]] == "Namespace", return the part. results of the comparison x.uri == y.uri

The comparision in 3(c) is does not verify that y is a namespace value.

1 MB = Member body (enter the ISO 3166 two-letter country code, e.g. CN for China; comments from the ISO/CS editing unit are identified by **) 2 Type of comment: ge = general te = technical ed = editorial NOTE Columns 1, 2, 4, 5 are compulsory. page 6 of 13 ISO electronic balloting commenting template/version 2001-10 Template for comments and response by the Editor on the comments submitted Date: 2005/01/27 Document: ISO/IEC DIS 22537

1 2 (3) 4 5 (6) (7)

Ecma 11.3.1 te The delete Operator Throw a TypeError exception of the operand of the Accept delete operator is if type XMLList. This section does not describe how delete x..@a works, for example, because the Descendants accessor, 11.2.3, does not return a Reference type (because [[Descendants]] for XML and XMLList do not return References, and the final steps of the semantics in 11.2.3 just propagate those return values), and the 11.3.1 Overview lists cases only for Reference types where the base object is XML and XMLList.

Ecma-262 Edition 3, 11.4.1, The delete Operator, specifies that delete on a non-Reference type returns true (step 2), so delete x..@a does nothing except evaluate to true. Ecma 12.1 te default xml namespace In functions that define the default xml Accept namespace, initialize the default xml namespace Default namespace is scoped lexically, but not hoisted to to the current value of the global default xml the top of function bodies in the same way that var namespace, at the beginning of the function block. definitions are. This make it hard to use and hard to compile. Ecma 12.3 ed "NOTE The for-each-in statement behaves differently Replace “differently than” with “differently from” Accept than the for-in statement." Should use "differently from" or "other than".

Ecma 12.3 ed off-by-one (too great) step numbering in "The mechanics Change numbering to “6 and 6a” and “7 and 7a”, Accept of enumerating the properties (steps 7 and 7a in the first respectively. algorithm, steps 8 and 8a in the second) is implementation dependent."

Ecma 13.3.2 te QName called with zero arguments is not specified don't Use empty string Accept want localName == "undefined" Ecma 13.4.4.1 te XML.prototype.constructor cannot be accessed because Make reserved for future use Accept. Add note that says XML [[Get]] does not lookup properties as Object [[Get]] the constructor property can 1 MB = Member body (enter the ISO 3166 two-letter country code, e.g. CN for China; comments from the ISO/CS editing unit are identified by **) 2 Type of comment: ge = general te = technical ed = editorial NOTE Columns 1, 2, 4, 5 are compulsory. page 7 of 13 ISO electronic balloting commenting template/version 2001-10 Template for comments and response by the Editor on the comments submitted Date: 2005/01/27 Document: ISO/IEC DIS 22537

1 2 (3) 4 5 (6) (7)

does. Ditto for XMLList.prototype.constructor. This may be write only in practice. not be the case in the future, creating a potential compatibility issue if implementations or program give these properties values. Ecma 13.4.4.6 te XML.prototype.child Step 1(a) depends on the [[Get]] Change P to propertyName. Accept with modifications method of x returning an XMLList of all children, and 1(b) that: depends on [[Get]] of that XMLList, called with index P, Step 2 set temporary to result of [[Get]] returning a list containing that one child, with 1.b. Set temporary to result Add Step 3 to return temporary converted to of get [[TargetObject]] referring to x and [[TargetProperty]] XMLList. presumably being null. 1.c. If temporary is undefined

then return empty list But [[Get]] on an XMLList, 9.2.1.1, does not return a list containing the indexed child, given a property index P -- it delegates to Object [[Get]]. Per 9.2, getting an indexed property from an XMLList will return undefined if P >= x.length(), otherwise it will return just the indexed child, not wrapped in a XMLList.

This contradicts the wording in the Overview ("If P is a numeric index, the child method returns a list ...."), but not the example.

Ecma 13.4.4.31 ed Step 9 "QNames" misspelled as "Qnames". Replace “Qnames” with “QNames”. Accept

Ecma 13.5.4.4 te XMLList.prototype.child seems to be missing at least a Add a Step 3, return m Accept 'return m' step at the end.

Ecma 13.5.4.9 ed Step 2 typo: "Returnt" should be "Return". Replace “Returnt” with “Return” Accept

Ecma 13.5.4.16 te Steps 1 and 3 specify undefined return, not null as is Return null from XMLList.prototype.parent() in the Reject. Returning undefined done for XML.prototype.parent(), making them case that there is no parent. is the correct behavior unnecessarily inconsistent. Ecma General ge Conformance section is missing. In particular it is not Add a conformance section, and add to that Accept

1 MB = Member body (enter the ISO 3166 two-letter country code, e.g. CN for China; comments from the ISO/CS editing unit are identified by **) 2 Type of comment: ge = general te = technical ed = editorial NOTE Columns 1, 2, 4, 5 are compulsory. page 8 of 13 ISO electronic balloting commenting template/version 2001-10 Template for comments and response by the Editor on the comments submitted Date: 2005/01/27 Document: ISO/IEC DIS 22537

1 2 (3) 4 5 (6) (7)

clear how implementers can extend E4X. conformance section the constraint that implementeors may not add to the set of methods of XML.prototype and XMLList.prototype Ecma General ge ToXMLName spelled as ToXmlName Replace globally ToXmlName with ToXMLName Accept

1 MB = Member body (enter the ISO 3166 two-letter country code, e.g. CN for China; comments from the ISO/CS editing unit are identified by **) 2 Type of comment: ge = general te = technical ed = editorial NOTE Columns 1, 2, 4, 5 are compulsory. page 9 of 13 ISO electronic balloting commenting template/version 2001-10 Template for comments and response by the Editor on the comments submitted Date: 2005/01/27 Document: ISO/IEC DIS 22537

1 2 (3) 4 5 (6) (7)

US 8 te Introduction of XML markup at the lexical level is a Remove XML markup from the lexical level of the Rejected, see email significant deviation in the underlying structure of an standard. comment. Ask rex if he has ECMAScript program and is inconsistent with what is any more insight on this done in other similar languages. A new form of XML literal, which could be comment. multiline, could be added to the standard; Allowing XML markup at the lexical level allows the However, it must have a clearly defined format writing of straight XML text in an easier fashion as quotes with a beginning and end that does not change are not needed around the text and multiline text does not with the XML text being represented. Having such need concatenation at the end of each line. (The a literal would make processing the language example in 13.4.4.2 clearly shows this.) However, this much easier and error checking much better. change is a major shift in the lexical structure of the language and does not help in the writing of XML text with embedded data from variables. The "ToXML" operator provides sufficient functionality that XML markup does not need to be shoved into the lexical structure of the language. US general te The document should use "shall" instead of "must" to be Change all "must" verbs to "shall" when used to Agreed consistent with ISO/IEC phraseology. specify mandatory requirements; use "may" when used to specify permissiveness ("is permitted to") US 4 ed Normative references should be clause 2, definitions Agreed clause 3, conformance clause 4, and everything else clause 5 and onward. US 3 p 1 ed Change "must" ==> "shall" Agreed US 3 p 3 ed Change "is permitted to provide" ==> "may" Agreed US 5 ed Move Clause 5 to an informative annex. Reject, but marking as informative US 6 ed Move Clause 6 to an informative annex. Reject, but marking as informative US general te In the cases where "type" means "datatype", replace Define type in the definition "type" with "datatype" to be consistent with JTC1 section terminology.

1 MB = Member body (enter the ISO 3166 two-letter country code, e.g. CN for China; comments from the ISO/CS editing unit are identified by **) 2 Type of comment: ge = general te = technical ed = editorial NOTE Columns 1, 2, 4, 5 are compulsory. page 10 of 13 ISO electronic balloting commenting template/version 2001-10 Template for comments and response by the Editor on the comments submitted Date: 2005/01/27 Document: ISO/IEC DIS 22537

1 2 (3) 4 5 (6) (7)

US 10.4.1, ed Change "var l = new XMLList" ==> "var List = new Note: This change (changing "l" ==> "List") is Accept. E.g. general XMLList". The problem is that the lower case ell looks necessary in several places within the document. like a number 1 (one). So just spell out "List" instead of var list = new List "l". US 13.5.4.3 te What does "!!!1" mean? Is this a typo? Not an issue. Only in PDF US 14 te The sentence "An implementation may define behaviour Possibly the following was intended: Reject, the proposed wording other than throwing a TypeError exception for the ToXML requires that "If the ToXML function or the ToXMLList function function and ToXMLList function when they are called implementations throw an are called with an argument of type Object, then with an argument of type Object" does not make sense exception, which is not the the error thrown is implementation-defined. Note: because "an implementation may define behaviour" case. Adding reference to The implementation-defined behaviour may doesn't make sense. section 16 of edition 3 to include throwing the TypeError exception." clarify. Replace controversial wording with “an implementation may choose to provide behaviour”

1 MB = Member body (enter the ISO 3166 two-letter country code, e.g. CN for China; comments from the ISO/CS editing unit are identified by **) 2 Type of comment: ge = general te = technical ed = editorial NOTE Columns 1, 2, 4, 5 are compulsory. page 11 of 13 ISO electronic balloting commenting template/version 2001-10 Template for comments and response by the Editor on the comments submitted Date: 2005/01/27 Document: ISO/IEC DIS 22537

1 2 (3) 4 5 (6) (7)

JP Overall ge Document style should be conforming to ISO/IEC Change the document style to conforming to Reject. Not required for first standard ISO/IEC standards edition of fast-track specifications JP Overall ge Clause for Definition of terminology is missing. Add a clause for Definition of Terms. Agreed JP 3 te Conformance definition given in Clause 3 is incomplete. Add a requirement of ISO/IEC 16262 to Clause 3. Agreed Currently, the conformance to ISO/IEC 16262 is not required, but obviously the conformance to E4X without the conformance to ISO/IEC 16262 is not possible. JP 4 Te Some references are not newest versions Change the references to the newest versions. Reject. Defer until edition 4 because this draft and implementations depend on earlier standards JP 4 Te References that are not actually cited should be moved to References that are not actually cited should be Agreed. If reference is in a Informative. moved to Notes. normative section then it goes in the normative references, otherwise it needs to be informative Add W3C XSLT as an informative references Change 4 to reference and add 4.1 normative and 4.2 informative JP 4 Te ISO/IEC 16262 must be referenced. Add a reference to ISO/IEC 16262:2001 Agreed. Change to ISO reference. JP 4 Te References to Unicode Technical Reports #8 and #15 Change the Unicode 2.0 reference to Unicode 4.0 Reject. Defer until edtion 4, should be deleted when referring to Unicode 4.0 is added. and delete the references to Unicode Technical because of dependency of Reports #8 and #15. edition 3 on Unicode 3 support JP 5 Ge This must be Informative. Declare informative, or move somewhere else. Agreed. Add sentence to

1 MB = Member body (enter the ISO 3166 two-letter country code, e.g. CN for China; comments from the ISO/CS editing unit are identified by **) 2 Type of comment: ge = general te = technical ed = editorial NOTE Columns 1, 2, 4, 5 are compulsory. page 12 of 13 ISO electronic balloting commenting template/version 2001-10 Template for comments and response by the Editor on the comments submitted Date: 2005/01/27 Document: ISO/IEC DIS 22537

1 2 (3) 4 5 (6) (7)

declare section as informative. JP 6 Ge This must be Informative. Declare informative, or move somewhere else. Agreed. Add sentence to declare as informative. JP 10.1.1 1st Te XML encoding handling must be specified. Add specific handling to the phrase, “ToString Remove text “an XML Paragraph returns an XML encoded string representing the encoded” so sentence reads entire XML object”. “ToString returns a string representing the entire XML object,” JP 14 ed Period missing - E4X extends the list of erros Add period. Agreed JP 14 ed Capitalization missing – implementations are not…. Capitalize as head of sentence. Agreed

1 MB = Member body (enter the ISO 3166 two-letter country code, e.g. CN for China; comments from the ISO/CS editing unit are identified by **) 2 Type of comment: ge = general te = technical ed = editorial NOTE Columns 1, 2, 4, 5 are compulsory. page 13 of 13 ISO electronic balloting commenting template/version 2001-10